Generating queries with AI
Don't feel like writing SQL? Describe what you want in plain language and the console writes the query for you, grounded in your snapshot's actual schema — real table and column names, not guesses.
How to use it
- In the Query Console, click Ask AI to open the AI bar.
- Describe what you want — for example: "accounts created this year with no opportunities" — and click Generate.
- The generated SQL lands in the editor for you to review.
- Run it, tweak it, or refine your request and regenerate.
[SCREENSHOT REQUIRED: AI query bar with a natural-language request and the generated SQL in the editor]
Pin an object with @-mentions
Type @ in the AI bar to mention a specific object — e.g. "biggest deals per region from @Opportunity". Mentioning an object pins its exact columns into the AI's context, which makes the generated query noticeably more precise on orgs with many similarly-named objects.
Example requests
- "contacts without an email, grouped by account"
- "monthly count of cases opened vs closed in 2026"
- "the 20 most recently modified
@Leadrecords with owner name" - "duplicate account names with how many times each appears"
Good to know
- Generated queries are read-only — only SELECT-style explorations are allowed; requests for anything else are refused.
- You always see the SQL before it runs; nothing executes without you.
- The AI sees your snapshot's schema (object and field names) to write correct queries — not your records.
- If AI generation isn't available in your environment, the bar says so; everything else in the console works normally.